embedding-ft

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt explicitly instructs setting and/or passing TIANGONG_LCA_API_KEY (including inline examples like TIANGONG_LCA_API_KEY="" and using --api-key), which encourages embedding API secrets into command lines or generated commands and therefore risks the model outputting secret values verbatim.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The wrapper runs the published CLI at runtime via "npm exec --yes --package=@tiangong-lca/cli@latest -- tiangong", which will fetch and execute remote code from the npm registry and is used by default unless TIANGONG_LCA_CLI_DIR is set, so @tiangong-lca/cli@latest is flagged.